Abstract: Aiming at the problem of running deviation which exists in the process of laser film slitting, this paper proposes a detection based on machine vision. Geometric features of laser film image are collected by using CCD camera, then it uses the improved Canny algorithm for edge detection of laser film after pretreatment and creates ROI area. Feature lines are obtained by using Hough transform and the deviation angle of laser film is calculated at last. The results of the experiment show that the improved Canny algorithm can detect more complete and clear edge details and it has strong adaptability. The effect of Hough transform for feature extraction on laser film is better, which provides a more accurate basis for the calculation offset of laser film in the future.

Key words: laser film, Hough transform, Canny algorithm, machine vision

摘要： 针对镭射膜纵切过程中的跑偏问题，提出一种基于机器视觉的检测方法。用CCD摄像机采集具有几何特征的镭射膜图像，对预处理后的图像采用改进Canny算法边缘检测并创建ROI区域，利用Hough变换得到特征直线，调用相关算法并计算出镭射膜跑偏角度。实验结果表明，改进Canny算法能够检测到较为完整的边缘细节，具备较强的自适应性，Hough变换对镭射膜的特征提取效果良好，为计算跑偏角度提供了较为准确的依据。

关键词: 镭射膜, Hough变换, Canny算法, 机器视觉